Allometry of post-flight cooling rates in moths: a comparison with vertebrate homeotherms

No homeothermic vertebrates weigh less than 2–3 g. Moreover, except for shrews, most mammsls and birds weighing less than 5 g are heterothermic and maintain high and regulated body temperatures only when they are active or when food is available in excess (see Bartholomew, 1972 for review). During periods of inactivity heterotherms have the capacity to allow body temperature to fall to within a degree or so of ambient temperature. The term heterothermy is ordinarily used with reference to hummingbirds, small bats, and small rodents. However, it also is applicable to those insects which maintain high and regulated body temperatures, when active, despite the fact that some of them weigh much less than a gram (see Heinrich, 1974 for a detailed review). Among these heterothermic insects are moths of the families fcphingidae and Saturniidae. Sphingids and saturniids differ in several physiological and morphological attributes that affect insect body temperatures; for example, relative size of the thorax and abdomen, wing loading, scalation, and feeding habits.

For endothermic animals to regulate body temperature, rates of heat production and heat loss must be balanced. Rates of heat loss of endotherms can be quantified in terms of cooling rates if the animals are dead or in a heterothermic state. The cooling rates of birds and mammals (Herreid & Kessel, 1967) when allometrically scaled show the same inverse proportionality as does the scaling of rates of heat production below the zone of thermal neutrality (Lasiewski, Weathers & Bernstein, 1967). Both heat loss and heat production below the zone of thermal neutrality can be expressed as weight-specific thermal conductance, cal (g h°C)^-1, where°C refers to the difference between body temperature and ambient temperature (a quantity that is often called the ‘temperature excess’ by students of insects).

Analysis of post-flight cooling rates in sphingids indicates that their weight-specific thermal conductance scaled against body weight shows the same slope as that of birds and mammals (Bartholomew & Epting, 1975). This indicates that the attainment of elevated and regulated flight temperatures by sphingids is primarily dependent on the high level of heat production by flight muscles rather than on especially reduced rates of heat loss.

Information on post-flight cooling in saturniids is limited to the cecropia moth Hyalophora cecropia (Oosthuizen, 1939; Heath & Adams, 1967; Hanegan & Heath, 1970; Bartholomew & Casey, 1973). Comparative data on body temperatures in both sphingids and saturniids during flight is available for many Central American species (Bartholomew & Heinrich, 1973). The present study examines the rates of post-flight cooling of saturniids ranging in weight from 81 to 2650 mg under controlled conditions of air temperature and air movement, and compares the allometry of their cooling rates with that of sphingids, birds and mammals.

The experimental work was carried out at the Tropical Training and Research Center of the Inter-American Institute of Agricultural Sciences of the Organization of American States in Turrialba, Costa Rica. The Center is 9° north of the equator at an elevation of 600 m and receives heavy rainfall at all times of the year.

The moths were captured at a light and handled as previously described for sphingids (Bartholomew & Epting, 1975). Each moth was anaesthetized, and a 40 gauge copper-constantan thermocouple was implanted in its thorax. Some of the smaller individuals initiated pre-flight warm-up as soon as the effects of the anesthesia wore off, but most of the moths remained quiescent until stimulated by gently touching them or blowing on them. The larger individuals aroused most readily in the late afternoon or early evening. Temperatures were measured to the nearest 0·1°C with a Microprobe thermometer (Bailey Instruments, Saddlebrook, New Jersey, U.S.A.).

Thoracic temperatures were recorded during warm-up and post-flight cooling in still air (ambient temperature 24–26°C). Each moth was then killed and weighed to the nearest mg. After our return to the University of California the moths were relaxed and pinned with the fore-and hind-wings overlapped in approximately the position used during flight. Each animal was photographed and wing area was measured with a planimeter on natural-size prints. The length and diameter of the thorax of each dried specimen were measured to the nearest 0·1 mm with calipers.

Data were processed at the UCLA Computer Center. Identifications were made by Julian P. Donahue of the Entomology Section of the Natural History Museum of Los Angeles County, where the moths are now on deposit.

After arousal, the moths warmed rapidly and tried to take off as soon as they approached flight temperature. Saturniids do not hover, and the cage was too small for free flight by any but the smallest individuals. Most of the animals flopped and bumbled about for a variable period (a few seconds to several minutes) before coming to rest and starting to cool down. We observed no effects of duration of attempted flight on the rate of cooling. Data similar to those in Fig. 1 were obtained for 25 moths of 8 genera belonging to 3 subfamilies (Table 1).

When ambient temperature is constant, the temperature of a passively-cooling object of homogeneous structure decreases exponentially and, therefore, describes a straight line when plotted on a semilogarithmic grid. When plotted thus, the cooling rates of only two of the saturniids are adequately described by straight lines. In all other cases the cooling rates were steeper initially than terminally (Fig. 1 and Table 1). From this we infer that in most cases cooling was physiologically-facilitated, initially, and passive, terminally. For the saturniid sample as a whole these differences were less pronounced than in sphingids (Bartholomew & Epting, 1975). For purposes of analysis and comparison we shall consider only the rates of cooling during the terminal 5 °C, all of which are linear when plotted semilogarithmically.

In saturniids the log-transformed values of terminal cooling rates decreased linearly with the log-transformed values of total body weight with a slope of —0·32 (Fig. 2). Most of the heat generated by the flight muscles of saturniids (Hanegan & Heath, 1970; Bartholomew & Casey, 1973; Hanegan, 1973) and sphingids (Heinrich 1970) is retained in the thorax while the abdomen remains at or near ambient temperature. Therefore, when considering the effects of size on rates of heat loss, it is appropriate to deal with the weight of the thorax rather than that of the total body. Since we could not determine the thoracic weights of the individuals on which we measured cooling rates without destroying them, we have used the calculated volume of their thoraxes as an index to thoracic weight. We treated the thorax as a cylinder and calculated its volume from measurements of length and diameter. The slope of the regression of the log-transformed values of thorax volume and body weight is approximately 0·8 (Fig. 3), which indicates that big saturniids have relatively smaller thoraxes than little ones. The terminal cooling rates were inversely related to thorax volume; the slope of the log-transformed values of terminal cooling rate on thoracic volume for saturniids was approximately —0·28 (Fig. 4).

We also captured and dissected a few saturniids and sphingids to obtain the weight of the thorax (not including appendages) as well as total body weight (Table 2). The regression of the proportional weight of the thorax against total weight was not significant in either family, P > 0·10. This finding agrees with our regression of thoracic volumes on body weight for sphingids (Fig. 3), which predicts that the proportional weight of the thorax should be independent of total body weight. The regression of thoracic volume on body weight for saturniids, however, predicts that larger saturniids should have a smaller proportion of body weight as thorax. The failure to find a significant regression of the proportional weight of the thorax against body weight for saturniids may be due to small sample size. Further data on thoracic volume and thoracic weights are necessary to clarify this point.

The patterns of post-flight cooling in saturniids although fundamentally similar to those of sphingids (Bartholomew & Epting, 1975) show differences that can be related to body proportions, insulation, and heat transfer to the abdomen.

When the terminal cooling rates are plotted against thoracic volume rather than against body weight, the effects of the differences in body proportions of sphingids and saturniids are eliminated. Because of the difference in body proportions (Fig. 3) in the two families, Fig. 2 overestimates the dependence of cooling on size in saturniids but not in sphingids. A comparison of the slopes of the log-transformed values of terminal cooling rates on thoracic volumes shows that cooling rate is less dependent on thoracic volume in saturniids than in sphingids (Fig. 4).

The positions of the regression lines in Fig. 4 show that in our samples, saturniids generally cool more slowly than sphingids, and are therefore, better insulated. Although quantitative data are lacking, saturniids have conspicuously long scales. Church (1960) showed that scales of moths have a pronounced effect on the retention of heat in the thorax. These two facts are consistent with our observation that even a very small (59 mg) satumiid (Hylesia sp.) was capable of elevating thoracic temperature 10 ° above air temperature.

In sphingids we found that cooling rates were significantly greater initially than terminally and interpreted this as evidence for facilitated transfer of heat to the abdomen during the period immediately after cessation of flight (Bartholomew & Epting, 1975). In our sample of saturniids the difference between initial and terminal cooling rates were less pronounced. More extensive sampling of saturniids will be necessary to confirm this difference between the two taxa. However, if real it may be correlated with the fact that saturniids as a group have somewhat lower flight temperatures than sphingids (Bartholomew & Heinrich, 1973) and fly only a few times during their life cycle. This difference does not affect the discussion which follows because the analyses are based entirely on terminal cooling rates.

As previously noted (Bartholomew & Epting, 1975) the weight-specific conductance of sphingids has the same slope as that of birds and mammals, but is 60 % greater in the range in which the weights of sphingids and homeothermic vertebrates overlap. Because of the high weight-specific conductance (based on total weight) of sphingids, we concluded that their elevated body temperatures result primarily from high rates, of heat production and are only secondarily dependent on effectiveness of insulation The weight-specific conductance of saturniids is even higher than that of sphingids (Fig. 5). Therefore, if we think in terms of total body weight, the same conclusion is also justified for saturniids. However, as noted by Heinrich & Bartholomew (1971) with warm-up rates in a variety of taxa, body weight is an unsatisfactory parameter for evaluating the achievement of elevated body temperature in heterothermic insects. This is due to the fact that heat generation occurs in the flight muscles and nearly all the heat is retained in the thorax while the abdomen typically remains at or near air temperature (Heinrich, 1974). No such simple compartmentalization occurs in bird and mammals. The fraction of the total body that heterothermic and homeothermic vertebrates maintain at core temperature obviously varies with body size, body proportions, and the difference between core and ambient temperatures.

The lines for vertebrates in Fig. 5 encompass the complex partitioning of heat exchanges involved in the maintenance of uniform core temperatures because they are based on measurements of oxygen consumption as well as on cooling rates ; moreover, they closely approximate a similar regression for birds calculated by Lasiewski, et al. (1967) that is based exclusively on oxygen consumption. However, the lines for saturniids and sphingids ignore the obvious partitioning of the body into two thermal components, the endothermic thorax and the abdomen. From the data in Table 2 and Figs. 3 and 5 it is possible to take this partitioning into account and to estimate the dependence of thermal conductance on thoracic weight rather than total body weight. Our data on thorax weight and volume in sphingids strongly indicate that the relative size of thorax is independent of total weight. The data on thoracic weight and volume in saturniids do not agree on this point probably because of the small sample of thoracic weights. In the absence of information about the specific gravity of the thorax in saturniids, we will use the slope of the regression of thorax volume on total weight in both families of moths for further analysis.

Herreid & Kessel (1967) found that the regressions of conductance on weight in birds and mammals do not differ significantly. Therefore, for purposes of comparison we have averaged the slopes and intercepts for the bird and mammal regressions and have used the new line to represent all endothermic amniotes.

In our sample of saturniids (Table 2) most of the body weights cluster near 0 ·4 g, and thoracic weights average about 35 % of total body weight. To derive the regression of weight-specific conductance on thoracic weight in saturniids it is necessary to transform both slope and intercept for the changes in thorac volume with increasing size. This slope was transformed by converting the scaling constant for weightspecific conductance (– 0 ·324) to a scaling constant for total conductance (0 ·676) and then dividing this quantity by the scaling constant for thoracic weight (1 ·165). The thoracic weight scaling constant was calculated using the scaling constants for conductance versus thoracic volume (–0 ·279) and conductance versus body weight (– 0 ·325) from Figs. 2 and 4. When the former is exponentiated to 1 ·0, the value for the scaling constant of thoracic weight is 1 ·165. The intercept was then shifted so that the new regression passes through the thoracic weight corresponding to a moth with a total body weight of 0 ·4 g. In sphingids thoracic weight averages 42 % of body weight and there is no dependence of the proportional volume of the thorax on body fc-eight. Consequently, it is only necessary to translate the regression line so that it passes through the thoracic weight corresponding to some arbitrary body weight, in this case, 1 g.

When these transformations are made the new regression lines for the moths correspond closely to the line for vertebrates. The weight-relative conductances of the thoraxes of sphingids and saturniids are essentially predicted by the regression line for vertebrates (Fig. 6). Consequently, the achieved levels of heat production in either of these moth families need not be as high as predicted on the basis of the weight-relative conductance of their entire bodies.

The similarity of the heterothermy of large insects with that of small birds and small mammals is a remarkable instance of evolutionary convergence. Despite the profound differences in morphology and physiological organization of insects and vertebrates, those members of both groups that depend on internally generated heat for the production of elevated and regulated body temperatures have had to satisfy the same physical constraint: the balancing of the rates of heat loss with rates of heat production. Thus, it is noteworthy that the rates of heat loss of sphingids and saturniids show the same allometric scaling as that of birds and mammals. From this we infer that the insulation of these moths is no more effective for animals of their size than is that of birds and mammals. Thus, the high body temperatures of satur-niids and sphingids during activity appear to depend primarily on high levels of heat production - a function for which the flight muscles of insects are uniquely well equipped because of their extremely high metabolic rates.

This study was supported by a grant (GB-32947X) from the U.S. National Science Foundation and carried out with the co-operation of the Inter-American Institute of Agricultural Science of the OAS and the Organization for Tropical Studies. We are grateful to Mr Arnold L. Erickson of the IICA and Sr Jorge R. Campabadal of OTS for allowing us to use the facilities of their respective institutions.